Every growing business hits the same question, and almost nobody answers it with evidence.
Where next?
It gets decided in a meeting, out loud, by whoever sounds most confident. Big city. Good income. Sounds like our kind of people. Let's try it.
Then it works or it doesn't, and nobody can say why.
My client runs local leagues — people sign up in their own town, get matched with players nearby, and play a season against their neighbors. It was growing fast, in a lot of places at once, and two lists were piling up.
Inbound: people writing in from all over, asking to bring a league to their town.
Outbound: markets they wanted to go after on purpose.
Both long. Neither ranked. And the instinct in the room was the instinct everybody has — go where the people and the money are.
So before ranking anything, I asked the only question that matters first.
What actually happened last season?
Not revenue. Signups. Revenue follows the count, and the count is what tells you whether a place caught on.
The answer embarrassed the assumption.
The strongest markets were mid-sized cities. The biggest, richest, most sprawling metros — the ones any strategy deck puts on slide one — underperformed badly.
Once you see why, you can't unsee it. A league only works if the people in it can physically get to each other. Scatter a handful of players across a giant metro and they will never play a single match. Put the same number in a town with edges and they will.
So that's what the model rewards, weighted by what the signups actually did, not by what sounded right in a meeting:
Density. Enough players close enough together that matches are practical.
Community stability. A proxy for the thing you can't measure directly — clubs, HOAs, groups that already know each other.
Looks like a place that worked. Reward markets that resemble the ones that took off.
A clear boundary. A town you can name, not “greater metro anything.”
A penalty for sprawl. Too big is a defect, not an asset. That one line is the whole thesis.
Size, barely. You need enough people. Bigger is not better.
And income?
Income came in dead last.
That's the finding. Every gut-feel version of this decision opens with “affluent.” The signups barely noticed.
Then three files, one for each person who had to do something with it:
The model — every current market ranked, every inbound request scored, fifty new markets recommended.
An outbound list — the ten markets to go after first, then the full fifty behind it. Sorted, ready to work.
An inbound list — everyone who'd written in, ranked by market strength and how many local contacts they had, so the best conversations happen first.
Here's the part I couldn't write three months ago.
The season that opened this month includes markets that came off that list.
The model didn't just rank things. It picked where to go. They went. Those leagues are live right now, full of people who are playing because a spreadsheet noticed their town was the right shape.
Which is a tidier ending than the truth, so — three honest notes.
One. The model carries a warning in its own summary tab, and I put it there on purpose. Census data does not fully explain who signs up. A great local organizer with a real list of people beats a better-scoring market every time.
Two. So three columns in the sheet are filled in by hand, not by data: is there someone here who could run this, is this community tight, is there a league nearby already. Human judgment gets its own columns, sitting right next to the census numbers, carrying real weight in the score.
Three. Some rows are honest about being guesses. A few of these places aren't markets the census actually tracks, so the sheet records which stand-in it used and how confident it is. A model that hides its soft spots is worse than one that admits them.
The takeaway isn't “score your markets.” It's two things.
First: the thing you're optimizing for is probably not the thing you assume. They'd have optimized for size and income. The data said proximity and stability. You only find that out by looking at what already happened — not by reasoning about what ought to work.
Second: a score doesn't make the decision, it shortens the list. This one turned two unranked piles into ten places worth a phone call. A human made every call after that. The markets that opened are the ones where a human found the right person.
So here's the one thing worth doing this week. Take your last twenty customers, locations, or accounts — whatever your version is — with a number next to each showing how well it went. Then open ChatGPT or Claude and paste this:
Here are my last 20 [customers/markets/accounts] and a number showing how each one performed: [paste]. What do the top five have in common that the bottom five don't? Rank those factors by how strongly they actually separate the two groups. Then tell me which factor I probably assume matters most, and where it really lands.
That last sentence is the one that earns its keep.
The gap between what you assume and what your data already knows is where the next good decision is hiding.
— Sarah
Reply 1 if you've been picking your next market by gut.
Reply 2 if you want the twenty-minute call to figure out what your data already knows.
